Coins Vibe / Knowledge library · Research · 2 min read
A metric needs a counting rule
Onchain data is public, but the meaning of an aggregate depends on definitions. Active addresses, transactions, transfer value, and fees measure different activity. A chart without methodology can make a precise-looking number answer a question it was never designed to measure.
An address is not a person. One user can create many addresses; one exchange address can serve many customers. Contracts, automated systems, and internal movements add more ambiguity. Describe the actual unit counted before translating an increase into a claim about adoption.
A hypothetical activity spike
A network’s daily active addresses double after an incentive campaign. Researchers should ask whether new addresses performed meaningful repeat activity, whether contracts created them automatically, and whether the pattern persists after rewards change. The spike is an observation; durable user growth is a further hypothesis.
Imagine transfer value also rises because a custodian reorganizes holdings between its own wallets. Gross onchain movement increases without an equivalent increase in outside economic payments. An adjusted transfer metric may try to remove some such activity, but the adjustment introduces classifications that require their own documentation.
Similarly, total value locked can rise when asset prices rise even if deposited token quantities remain unchanged. Comparing value in dollars and underlying units helps separate market repricing from net deposits. Cross-protocol positions can also create overlapping claims that should not be casually added as independent capital.
Compare compatible periods and systems
Daily metrics need a timezone and boundary convention. Rolling windows differ from calendar days. Recent blocks may be incomplete, and data providers may revise results after reorganization handling or label changes. Record retrieval time and any completeness limits.
Cross-chain comparisons require additional care. A transaction on one system may bundle many user actions, while another represents those actions separately. Fee totals can reflect execution demand, token price, and fee-market design. Raw counts are not automatic scoreboards of usefulness.
A good research note states both a claim and its plausible alternatives. For example: “fees increased; possible explanations include higher activity, congestion, changed pricing, or more expensive transaction types.” Then choose additional observations capable of distinguishing those explanations.
A metric checklist
- Write the question, unit, counting rule, network, and time window.
- Distinguish addresses from users and gross movements from economic payments.
- Compare value-based metrics with underlying asset quantities where relevant.
- Check adjustment heuristics, exclusions, and revision policy.
- Use a second independent measure to test the proposed explanation.
Make a small reproducible sample before trusting a large dashboard. Inspect several underlying transactions that contribute to the metric and several that are excluded. This reveals whether the definition fits your question and helps explain discrepancies between providers without assuming one must be wrong.
Frequently asked questions
Are onchain metrics objective?
Raw records are observable, but aggregation and interpretation involve choices. Reproducible methodology makes those choices inspectable; it does not remove them.
Does more transaction activity mean more investment value?
No. Activity can be automated, subsidized, internal, or economically small. Token value requires additional reasoning about rights, demand, supply, and market conditions.
